System and method for operant learning brain machine interface by receiving electrodes and neural signals at neural processor

Abstract

A method, system and computer readable media for a BMI using a fixed decoder based on ratios of different frequency bands, making the decoder robust, less jittery, and resistant to artifacts. The fixed decoder can be configured to use a limited subset of available channels. The decoder can therefore be optimized for each human subject (frequency bands to use, ratios to process the received signals, which channels, weights, etc.) and then fixed. Output from the fixed decoder can be provided to a training program that implements specific feedback and training parameters, thereby enabling subjects to learn to control devices rapidly, as well as consolidate this control. The training program provides continuous feedback of the current transformation being output by the fixed decoder in conjunction with feedback of the past transformations (e.g., up to a second before) and saliency of the feedback when goals of the task are achieved.